FIELD

[0001]
The present invention relates to coding and decoding of a sound signal, for example an audio signal. More specifically, but not exclusively, the present invention relates to variable bit rate LPC (Linear Prediction Coefficients) filter quantizing and inverse quantizing device and method.
BACKGROUND

[0002]
The demand for efficient digital speech and audio coding techniques with a good tradeoff between subjective quality and bit rate is increasing in various application areas such as teleconferencing, multimedia, and wireless communication.

[0003]
A speech coder converts a speech signal into a digital bit stream which is transmitted over a communication channel or stored in a storage medium. The speech signal to be coded is digitized, that is sampled and quantized using for example 16bits per sample. A challenge of the speech coder is to represent the digital samples with a smaller number of bits while maintaining a good subjective speech quality. A speech decoder or synthesizer converts the transmitted or stored bit stream back to a speech signal.

[0004]
CodeExcited Linear Prediction (CELP) coding is one of the best techniques for achieving a good compromise between subjective quality and bit rate. The CELP coding technique is a basis for several speech coding standards both in wireless and wireline applications. In CELP coding, the speech signal is sampled and processed in successive blocks of L samples usually called frames, where L is a predetermined number of samples corresponding typically to 1030 ms of speech. A linear prediction LP) filter is computed and transmitted every frame; the LP filter is also known as LPC (Linear Prediction Coefficients) filter. The computation of the LPC filter typically uses a lookahead, for example a 515 ms speech segment from the subsequent frame. The Lsample frame is divided into smaller blocks called subframes. In each subframe, an excitation signal is usually obtained from two components, a past excitation and an innovative, fixedcodebook excitation. The past excitation is often referred to as the adaptivecodebook or pitchcodebook excitation. The parameters characterizing the excitation signal are coded and transmitted to the decoder, where the excitation signal is reconstructed and used as the input of the LPC filter.

[0005]
In applications such as multimedia streaming and broadcast, it may be required to encode speech, music, and mixed content at low bit rate. For that purpose, encoding models have been developed which combine a CELP coding optimized for speech signals with transform coding optimized for audio signals. An example of such models is the AMRWB+ [1], which switches between CELP and TCX (Transform Coded excitation). In order to improve the quality of music and mixed content, a long delay is used to allow for finer frequency resolution in the transform domain. In AMRWB+, a socalled superframe is used which consists of four CELP frames (typically 80 ms).

[0006]
A drawback is that, although the CELP coding parameters are transmitted once every 4 frames in AMRWB+, quantization of the LPC filter is performed separately in each frame. Also, the LPC filter is quantized with a fixed number of bits per frame in the case of CELP frames.
BRIEF DESCRIPTION OF THE DRAWINGS

[0007]
In the appended drawings:

[0008]
FIG. 1 is a block diagram illustrating an absolute and multireference differential LPC filter quantizer and quantizing method;

[0009]
FIG. 2 is a schematic diagram illustrating an openloop quantization scheme;

[0010]
FIG. 3 is a flow chart illustrating a device and method for determining LPC filters to be transmitted in a configuration in which four (4) LPC filters are used and transmitted in a superframe;

[0011]
FIG. 4 a is a typical LPC analysis window and typical LPC analysis center position when one LPC filter is estimated per frame (or superframe) in an LPCbased codec, wherein LPC0 corresponds to a last LPC filter of the previous frame (or superframe);

[0012]
FIG. 4 b is a typical LPC analysis window when four (4) LPC filters are estimated per frame (or superframe) in an LPCbased codec, wherein the LPC analysis window is centered at the end of the frame;

[0013]
FIG. 5 is a flow chart illustrating an example of an outoftheloop quantization scheme;

[0014]
FIG. 6 is a schematic block diagram of a weighted algebraic LPC quantizer and quantizing method;

[0015]
FIG. 7 is a schematic block diagram of a weighted algebraic LPC inverse quantizer and quantizing method;

[0016]
FIG. 8 is a schematic block diagram of a quantizer and quantizing method; and

[0017]
FIG. 9 is a schematic block diagram of a decoder and decoding method.
DETAILED DESCRIPTION

[0018]
According to nonrestrictive illustrative embodiments of the present invention, there are provided:

[0019]
A device for quantizing a LPC filter in the form of an input vector in a quantization domain, comprising: means for computing a firststage approximation of the input vector; means for subtracting the firststage approximation from the input vector to produce a residual vector; means for calculating a weighting function from the firststage approximation; means for applying the weighting function to the residual vector; and means for quantizing the weighted residual vector to supply a quantized weighted residual vector.

[0020]
A device for quantizing a LPC filter in the form of an input vector in a quantization domain, comprising: a calculator of a firststage approximation of the input vector; a subtractor of the firststage approximation from the input vector to produce a residual vector; a calculator of a weighting function from the firststage approximation; a warper of the residual vector with the weighting function; and a quantizer of the weighted residual vector to supply a quantized weighted residual vector.

[0021]
A device for quantizing a LPC filter in the form of an input vector in a quantization domain, comprising: a calculator of a firststage approximation of the input vector; a subtractor of the firststage approximation from the input vector to produce a residual vector; a calculator of a weighting function from the firststage approximation; a warper of the residual vector with the weighting function; and a quantizer of the weighted residual vector to supply a quantized weighted residual vector. The calculator of the firststage approximation is selected from the group consisting of: an absolute quantizer of the input vector; a quantizer of a previous LPC vector; a quantizer of a future LPC vector; and an interpolator of previous quantized and/or future quantized LPC vectors; to give an estimate of the input vector.

[0022]
A device for quantizing a LPC filter in the form of an input vector in a quantization domain, comprising: a calculator of a firststage approximation of the input vector; a subtractor of the firststage approximation from the input vector to produce a residual vector; a calculator of a weighting function from the firststage approximation; a warper of the residual vector with the weighting function; and a quantizer of the weighted residual vector to supply a quantized weighted residual vector. The calculator of the weighting function calculates different weights, and the warper applies the different weights to components of the residual vector.

[0023]
A device for quantizing a LPC filter in the form of an input vector in a quantization domain, comprising: a calculator of a firststage approximation of the input vector; a subtractor of the firststage approximation from the input vector to produce a residual vector; a calculator of a weighting function from the firststage approximation; a warper of the residual vector with the weighting function; and a quantizer of the weighted residual vector to supply a quantized weighted residual vector. The quantizer of the weighted residual vector comprises a variable bit rate quantized.

[0024]
A device for inverse quantizing of a LPC filter, comprising: means for receiving coded indices representative of a firststage approximation of a vector representative of the LPC filter in a quantization domain, and a quantized weighted residual version of the vector; means for calculating an inverse weighting function from the firststage approximation; means for inverse quantizing the quantized weighted residual version of the vector to produce a weighted residual vector; means for applying the inverse weighting function to the weighted residual vector to produce a residual vector; and means for adding the firststage approximation with the residual vector to produce the vector representative of the LPC filter in the quantization domain.

[0025]
A device for inverse quantizing of a LPC filter, comprising: means for receiving coded indices representative of a firststage approximation of a vector representative of the LPC filter in a quantization domain, and a quantized weighted residual version of the vector; a calculator of an inverse weighting function from the firststage approximation; an inverse quantizer of the quantized weighted residual version of the vector to produce a weighted residual vector; a multiplier of the weighted residual vector by the inverse weighting function to produce a residual vector; and an adder of the firststage approximation with the residual vector to produce the vector representative of the LPC filter in the quantization domain.

[0026]
A device for inverse quantizing of a LPC filter, comprising: means for receiving coded indices representative of a firststage approximation of a vector representative of the LPC filter in a quantization domain, and a quantized weighted residual version of the vector; a calculator of an inverse weighting function from the firststage approximation, an inverse quantizer of the quantized weighted residual version of the vector to produce a weighted residual vector; a multiplier of the weighted residual vector by the inverse weighting function to produce a residual vector; and an adder of the firststage approximation with the residual vector to produce the vector representative of the LPC filter in the quantization domain. The received coded indices include an index representative of a quantization mode.

[0027]
A device for inverse quantizing of a LPC filter, comprising: means for receiving coded indices representative of a firststage approximation of a vector representative of the LPC filter in a quantization domain, and a quantized weighted residual version of the vector; a calculator of an inverse weighting function from the firststage approximation; an inverse quantizer of the quantized weighted residual version of the vector to produce a weighted residual vector; a multiplier of the weighted residual vector by the inverse weighting function to produce a residual vector; and an adder of the firststage approximation with the residual vector to produce the vector representative of the LPC filter in the quantization domain. The inverse quantizer of the quantized weighted residual version of the vector comprises a variable bit rate inverse quantizer.

[0028]
A method for quantizing a LPC filter in the form of an input vector in a quantization domain, comprising: computing a firststage approximation of the input vector; subtracting the firststage approximation from the input vector to produce a residual vector; calculating a weighting function from the firststage approximation; applying the weighting function to the residual vector; and quantizing the weighted residual vector to supply a quantized weighted residual vector.

[0029]
A method for quantizing a LPC filter in the form of an input vector in a quantization domain, comprising: computing a firststage approximation of the input vector; subtracting the firststage approximation from the input vector to produce a residual vector; calculating a weighting function from the firststage approximation; applying the weighting function to the residual vector; and quantizing the weighted residual vector to supply a quantized weighted residual vector. Computing the firststage approximation is selected from the group consisting of: absolute quantizing of the input vector; quantizing of a previous LPC vector; quantizing of a future LPC vector; and interpolating of previous quantized and/or future quantized LPC vectors to give an estimate of the input vector.

[0030]
A method for quantizing a LPC filter in the form of an input vector in a quantization domain, comprising: computing a firststage approximation of the input vector; subtracting the firststage approximation from the input vector to produce a residual vector; calculating a weighting function from the firststage approximation; applying the weighting function to the residual vector; and quantizing the weighted residual vector to supply a quantized weighted residual vector. Calculating a weighting function comprises calculating different weights, and applying the weighting function comprises applying the different weights to components of the residual vector.

[0031]
A method for quantizing a LPC filter in the form of an input vector in a quantization domain, comprising: computing a firststage approximation of the input vector; subtracting the firststage approximation from the input vector to produce a residual vector; calculating a weighting function from the firststage approximation; applying the weighting function to the residual vector; and quantizing the weighted residual vector to supply a quantized weighted residual vector. Quantizing the weighted residual vector comprises using a variable bit rate quantizer.

[0032]
A method for inverse quantizing of a LPC filter, comprising: receiving coded indices representative of a firststage approximation of a vector representative of the LPC filter in a quantization domain, and a quantized weighted residual version of the vector; calculating an inverse weighting function from the firststage approximation; inverse quantizing the quantized weighted residual version of the vector to produce a weighted residual vector; applying the inverse weighting function to the weighted residual vector to produce a residual vector; and adding the firststage approximation with the residual vector to produce the vector representative of the LPC filter in the quantization domain.

[0033]
A method for inverse quantizing of a LPC filter, comprising: receiving coded indices representative of a firststage approximation of a vector representative of the LPC filter in a quantization domain, and a quantized weighted residual version of the vector; calculating an inverse weighting function from the firststage approximation; inverse quantizing the quantized weighted residual version of the vector to produce a weighted residual vector; applying the inverse weighting function to the weighted residual vector to produce a residual vector; and adding the firststage approximation with the residual vector to produce the vector representative of the LPC filter in the quantization domain. The received coded indices include an index representative of a quantization mode.

[0034]
A method for inverse quantizing of a LPC filter, comprising: receiving coded indices representative of a firststage approximation of a vector representative of the LPC filter in a quantization domain, and a quantized weighted residual version of the vector; calculating an inverse weighting function from the firststage approximation; inverse quantizing the quantized weighted residual version of the vector to produce a weighted residual vector; applying the inverse weighting function to the weighted residual vector to produce a residual vector; and adding the firststage approximation with the residual vector to produce the vector representative of the LPC filter in the quantization domain. Inverse quantizing the quantized weighted residual version of the vector comprises variable bit rate inverse quantizing the quantized weighted residual version of the vector.

[0035]
The foregoing and other features of the present invention will become more apparent upon reading of the following nonrestrictive description of embodiments thereof, given by way of example only with reference to the accompanying drawings.

[0000]
Differential Quantization with a Choice of Possible References

[0036]
Differential quantization with a choice between several possible references is used. More specifically, a LPC filter is differentially quantized relative to several possible references.

[0037]
Consecutive LPC filters are known to exhibit a certain degree of correlation. To take advantage of this correlation, LPC quantizers generally make use of prediction. Instead of quantizing the vector of Linear Prediction Coefficients (LPC vector) representing the LPC filter directly, a differential (or predictive) quantizer first computes a predicted value of this LPC vector and, then, quantizes the difference (often called prediction residual) between the original LPC vector and the predicted LPC vector.

[0038]
Prediction is normally based on previous values of the LPC filter. Two types of predictors are commonly used: moving average (MA) and autoregressive (AR) predictors. Although AR predictors are often more efficient in reducing the L2norm (mean square) of the data to be quantized than MA predictors, the latter are sometimes useful because they are less prone to error propagation in case of transmission errors [2].

[0039]
Since the L2norm of the prediction residual is on average lower than the L2norm of the original LPC vector (the ratio between the two depending on the degree of predictability of the LPC filter), a differential (or predictive) quantizer can achieve the same degree of performance as an absolute quantizer but at a lower bit rate.

[0040]
On average, prediction is indeed efficient at reducing the L2norm of the data to be quantized. This behavior is not constant however; prediction is much more efficient during stable segments of signal than during transitional segments. Prediction can even lead to increased L2norm values when the LPC filters change fast. Some performance improvement can be achieved by considering two different predictors, one for highly predictive segments the other for less predictive segments [3, 4]. As mentioned in the foregoing description, this technique uses only past values of the LPC filter.

[0041]
To overcome this problem, it is proposed to differentially quantize a LPC filter relative to a reference, for example a reference filter, chosen among a number of possible references. The possible reference filters are already quantized past or future LPC filters (hence available at the decoder as at the encoder), or the results of various extrapolation or interpolation operations applied to already quantized past or future LPC filters. The reference filter that provides the lower distortion at a given rate, or the lower bit rate for a given target distortion level, is selected.

[0042]
FIG. 1 is a block diagram illustrating a multireference LPC filter quantization device and method. A given LPC filter 101 represented by a vector of Linear Prediction Coefficients is inputted to the multireference LPC filter quantization device and method. The input LPC filter 101 is differentially quantized with respect to a reference chosen among a number of possible references 1, 2, . . . , n. Possible references comprise:

[0043]
past or future quantized LPC filters;

[0044]
the result of extrapolation or interpolation operations applied to past or future quantized LPC filters; or

[0045]
any quantized value available both at the encoder and the decoder.

[0046]
As a nonlimitative example, the input LPC filter 101 can be differentially quantized with respect to the previous quantized LPC filter, the following quantized LPC filter, or a mean value of those two previous and following quantized LPC filters. A reference can also be a LPC filter quantized using an absolute quantizer, or the result of any kind of interpolation, extrapolation or prediction (AR or MA) applied to already quantized LPC filters.

[0047]
Operations 102 and 103 _{1}, 103 _{2}, . . . , 103 _{n}: Still referring to FIG. 1, the input LPC filter 101 is supplied to an absolute quantizer (Operation 102) and to differential quantizers (Operations 103 _{1}, 103 _{2}, . . . , 103 _{n}). The absolute quantizer (Operation 102) quantizes the absolute value (not a difference) of the input LPC filter 101. The differential quantizers (Operations 103 _{1}, 103 _{2}, . . . , 103 _{n}) are designed to differentially quantize the input LPC filter 101 with respect to respective references 1, 2, . . . n.

[0048]
Operation 104: The multireference LPC filter quantization device and method of FIG. 1 comprises a selector for selecting a reference amongst the references 1, 2, . . . , n that provides the lowest distortion level at a given bit rate, or the lowest bit rate for a given target distortion level. More specifically, the selector (Operation 104) uses a selection criterion that minimizes the bit rate to achieve a certain target level of distortion, or that minimizes the level of distortion produced at a given bit rate.

[0049]
In Operation 104, the selection of a reference amongst references 1, 2, . . . , n to be actually used in the differential quantization process can be performed in closedloop or in openloop.

[0050]
In closedloop, all possible references are tried and the reference that optimizes a certain criterion of distortion or bit rate is chosen. For example, the closedloop selection can be based on minimizing a weighted meansquared error between the input LPC vector and the quantized LCP vector corresponding to each reference. Also, the spectral distortion between the input LPC vector and the quantized LPC vector can be used. Alternatively, the quantization using the possible references can be performed while maintaining a distortion under a certain threshold, and the reference that both meets with this criterion and uses the smaller number of bits is chosen. As will be explained in the following description, a variable bit rate algebraic vector quantizer can be used to quantize the scaled residual vector (difference between the input LPC vector and the reference) which uses a certain bit budget based on the energy of the scaled residual vector. In this case, the reference which yields the smaller number of bits is chosen.

[0051]
In openloop, the selector of Operation 104 predetermines the reference based on the value of the Linear Prediction Coefficients of the input LPC filter to be quantized and of the Linear Prediction Coefficients of the available reference LPC filters. For example, the L2norm of the residual vector is computed for all references and the reference that yields the smaller value is chosen.

[0052]
Operation 105: Following the selection of one of the references 1, 2, . . . , n by the Operation 104, a transmitter (Operation 105) communicates or signals to the decoder (not shown) the quantized LPC filter (not shown) and an index indicative of the quantization mode (suboperation 1051), for example absolute or differential quantization. Also, when differential quantization is used, the transmitter (Operation 105) communicates or signals to the decoder indices representative of the selected reference and associated differential quantizer of Operations 103 _{1}, 103 _{2}, . . . , 103 _{n }(suboperation 105 _{2}). Some specific bits are transmitted to the decoder for such signaling.

[0053]
Using a number of different possible references makes differential quantization more efficient in terms of reduction of the L2norm of prediction residual compared to restricting to past values only as in conventional prediction. Also, for a given target level of distortion, this technique is more efficient in terms of average bit rate.
Switched Absolute or Differential Quantization

[0054]
According to a second aspect, switched absolute/differential (or predictive) quantization is used. FIG. 1 illustrates an example of an absolute/differential scheme which selects between one absolute quantizer (Operation 102) and n differential quantizers (Operations 103 _{1}, 103 _{2}, . . . , 103 _{n}) using respective, different references 1, 2, . . . , n. Again, the selection of a quantizer can be made by the selector of Operation 104 amongst the absolute and differential quantizers (Operations 102 and 103 _{1}, 103 _{2}, . . . , 103 _{n}), wherein the selected quantizer will, according to the selection criterion, minimize the level of distortion produced at a given bit rate or minimize the bit rate to achieve a target level of distortion.

[0055]
Some LPC filters can be coded using the absolute quantizer (Operation 102). The other LPC filters are coded differentially with respect to one or several reference LPC filters in the differential quantizers (Operations 103 _{1}, 103 _{2}, . . . , 103 _{n}).

[0056]
The absolute quantizer (Operation 102) can be used as a safetynet solution for the otherwise differentially quantized LPC filters, for example in the case of large LPC deviations or when the absolute quantizer (Operation 102) is more efficient than the differential quantizers (Operations 103 _{1}, 103 _{2}, . . . , 103 _{n}) in terms of bit rate. The reference LPC filter(s) can be all within the same superframe to avoid introducing dependencies between superframes which usually pose problems in case of transmission errors (packet losses or frame erasures).

[0057]
As explained in the foregoing description, the use of prediction in LPC quantization leads to a reduced L2norm of the data to be quantized and consequently to a reduction in average bit rate for achieving a certain level of performance. Prediction is not always equally efficient however. In switched LPC [3, 4], a preclassification of the LPC filter is performed and different predictors are used depending on the predictability of the LPC filter to be quantized. However this technique has been developed in the context of a fixed bit rate, the two differential quantizers requiring a same number of bits to code an LPC filter.

[0058]
Also, there may be provided one or several absolute quantizers (Operation 102). Moreover, there may be provided one or several differential (predictive) quantizers (Operations 103 _{1}, 103 _{2}, . . . , 103 _{n}). Several differential quantizers (Operations 103 _{1}, 103 _{2}, . . . , 103 _{n}) involves several possible references such as 1, 2, . . . , n and/or several differential quantizer sizes and/or structures.

[0059]
As described in the foregoing description, when several differential quantizers (Operations 103 _{1}, 103 _{2}, . . . , 103 _{n}) are used, selection of the actual differential quantizer to be used can be performed in an openloop or in a closedloop selecting process.

[0060]
When differential quantization fails to achieve a target level of distortion, or when absolute quantization requires a smaller number of bits than differential quantization to achieve that level of distortion, absolute quantization is used as a safetynet solution. One or several bits, depending on the number of possible absolute and differential quantizers is (are) transmitted through the transmitter (Operation 105) to indicate to the decoder (not shown) the actual quantizer being used.

[0061]
Absolute/differential quantization combines the advantages of predictive quantization (reduction in bit rate associated with the reduction of the L2norm of the data to be quantized) with the generality of absolute quantization (which is used as a safetynet in case differential (or predictive) quantization does not achieve a target, for example unnoticeable, level of distortion).

[0062]
When several differential quantizers (Operations 103 _{1}, 103 _{2}, . . . , 103 _{n}) are included, these differential quantizers can make use of either a same predictor or different predictors. In particular, but not exclusively, these several differential quantizers can use the same prediction coefficients or different prediction coefficients.

[0063]
The decoder comprises means for receiving and extracting from the bit stream, for example a demultiplexer, (a) the quantized LPC filter and (b) the index (indices) or information:

[0064]
about the quantization mode to determine if the LPC filter has been quantized using absolute quantization or differential quantization; and

[0065]
about the reference amongst the plurality of possible references that has been used for quantizing the LPC filter.

[0066]
If the information about the quantization mode indicates that the LPC filter has been quantized using absolute quantization, an absolute inverse quantizer (not shown) is provided for inverse quantizing the quantized LPC filter. If the information about the quantization mode indicates that the LPC filter has been quantized using differential quantization, a differential inverse quantizer (not shown) then differentially inverse quantizes the multireference differentially quantized LPC filter using the reference corresponding to the extracted reference information.
OutoftheLoop Quantization Scheme

[0067]
The AMRWB+ codec is a hybrid codec that switches between a timedomain coding model based on the ACELP coding scheme, and a transformdomain coding model called TCX. The AMRWB+proceeds as follows [1]:

[0068]
The input signal is segmented into superframes of four (4) frames;

[0069]
Each superframe is encoded using a combination of four (4) possible coding modes, each coding mode covering a different duration:

 ACELP (covering a duration of one (1) frame);
 TCX256 (covering a duration of one (1) frame);
 TCX512 (covering a duration of two (2) frames); and
 TCX1024 (covering a duration of four (4) frames).

[0074]
There are therefore 26 possible mode combinations to code each superframe.

[0075]
For a given superframe, the combination of modes which minimizes a total weighted error is determined by a “closedloop” mode selection procedure. More specifically, instead of testing the 26 combinations, the selection of the mode is performed through eleven (11) different trials (tree search, see Table 1). In AMRWB+ codec, the closedloop selection is based on minimizing the meansquared error between the input and codec signal in a weighted domain (or maximizing the signal to quantization noise ratio).

[0000]
TABLE 1 

The 11 trials for closedloop mode selection in AMRWB+ 
Trial 
Frame 1 
Frame 2 
Frame 3 
Frame 4 

1 
ACELP 



2 
TCX256 
3 

ACELP 
4 

TCX256 
5 
TCX512 



6 


ACELP 

7 


TGX256 
8 



ACELP 
9 



TCX256 
10 


TCX512 

11 
TCX1024 


[0076]
The LPC filters are one of the various parameters transmitted by the AMRWB+codec. Following are some key elements regarding the quantization and transmission of those LPC filters.

[0077]
Although the different coding modes do not cover the same number of frames, the number of LPC filters transmitted to the decoder is the same for all coding modes and equal to 1. Only the LPC filter corresponding to the end of the covered segment is transmitted. More specifically, in the case of TCX1024, one (1) LPC filter is calculated and transmitted for a duration of four (4) frames. In the case of TCX512, one (1) LPC filter is calculated and transmitted for a duration of two (2) frames. In the case of TCX256 or ACELP, one (1) LPC filter is calculated and transmitted for the duration of one (I) frame.

[0078]
The AMRWB+ codec uses a (first order movingaverage) predictive LPC quantizer. The operation of the latter quantizer depends on the previously transmitted LPC filter, and consequently on the previously selected coding mode. Therefore, because the exact combination of modes is not known until the entire superframe is coded, some LPC filters are encoded several times before the final combination of modes is determined.

[0079]
For example, the LPC filter located at the end of frame 3 is transmitted to the decoder only when the third frame is encoded as ACELP or TCX256. It is not transmitted when frames 3 and 4 are jointly encoded using TCX512. With regards to the LPC filter located at the end of frame 2, it is transmitted in all combinations of modes except in TCX1024. Therefore, the prediction performed when quantizing the last LPC filter of the superframe depends on the combination of modes for the whole superframe.

[0080]
The principle of the disclosed technique is that the order in which the LPC filters are quantized is chosen so that, once the closedloop decision is finalized, the quantization information corresponding to the unnecessary LPC filters can be skipped from the transmission with no effect on the way the other filters will be transmitted and decoded at the decoder. For each LPC filter to be quantized using the differential quantization strategy described above, this imposes some constraints on the possible reference LPC filters.

[0081]
The following example is given with reference to FIG. 2.

[0082]
Operation 1 of FIG. 2: To avoid any inter superframe dependencies, at least one LPC filter is quantized using an absolute LPC quantizer. Since filter LPC4 of frame 4 of the superframe is always transmitted whatever the coding mode combination determined by the closedloop selection procedure, it is convenient to quantize that filter LPC4 using an absolute quantizer.

[0083]
Operation 2 of FIG. 2: The next LPC filter to be quantized is filter LPC2 of frame 2 of the superframe which is transmitted for all combinations of modes except for TCX1024. A differential quantizer can be used, for example to code the difference between filter LPC2 and the absolute quantized version of filter LPC4. The same absolute quantizer as used for coding filter LPC4 can also be used as a safetynet solution, for example in the case of large LPC deviations or when the absolute LPC quantizer is more efficient than the differential quantizer in terms of bit rate and/or level of distortion.

[0084]
Operation 3 of FIG. 2: The remaining two LPC filters (filter LPC1 of frame 1 of the superframe and filter LPC3 of frame 3 of the superframe) are also quantized using the same differential/absolute quantization strategy. Both LPC filters can be quantized relative to the quantized version of filter LPC2. Some alternative strategies are given herein below.

[0085]
FIG. 5 is a flow chart illustrating in more detail an example of an outoftheloop quantization scheme.

[0086]
Operation 501: An absolute quantizer quantizes the filter LPC4.

[0087]
Operation 502: Operation 512 is optional and used in a first LPCbased coding frame after a nonLPCbased coding frame. An absolute quantizer quantizes the filter LPC0 or a differential quantizer differentially quantizes the filter LPC0 relative to the quantized filter LPC4. The filter LPC0 is the last LPC filter (LPC4) from the previous superframe and can be used as a possible reference for quantizing the filters LPC1 to LPC4.

[0088]
Operation 503: An absolute quantizer quantizes the filter LPC2 or a differential quantizer differentially quantizes the filter LPC2 relative to the quantized filter LPC4 used as reference.

[0089]
Operation 504: An absolute quantizer quantizes the filter LPC1, a differential quantizer differentially quantizes the filter LPC1 relative to the quantized filter LPC2 used as reference, or a differential quantizer differentially quantizes the filter LPC1 relative to (quantized filter LPC2+quantized filter LPC0)/2) used as reference.

[0090]
Operation 505: An absolute quantizer quantizes the filter LPC3, a differential quantizer differentially quantizes the filter LPC3 relative to the quantized filter LPC2 used as reference, a differential quantizer differentially quantizes the filter LPC3 relative to quantized filter LPC4 used as reference, or a differential quantizer differentially quantizes the filter LPC3 relative to (quantized filter LPC2+quantized filter LPC4)/2) used as reference.

[0091]
FIG. 3 is a flow chart illustrating determination of LPC filters to be transmitted in a configuration where four (4) LPC filters can be calculated and transmitted in a superframe.

[0092]
First of all it should be kept in mind that quantized filter LPC1 is transmitted only when ACELP and/or TCX256 is selected for the first half of the superframe. Similarly, filter LPC3 is transmitted only when ACELP and/or TCX256 is used for the second half of that superframe.

[0093]
Operation 301: Filter LPC1 of frame 1 of the superframe, filter LPC2 of frame 2 of the superframe, filter LPC3 of frame 3 of the superframe, and filter LPC4 of frame 4 of the superframe are quantized using for example the quantization strategy illustrated and described in relation to FIGS. 2 and 5. Of course, other quantization strategies are possible.

[0094]
Operation 302: Closedloop selection of the coding modes as described hereinabove is performed.

[0095]
Operation 303: The quantized filter LPC4 is transmitted to the decoder for example through the transmitter 105 of FIG. 1. The decoder comprises:

[0096]
means for receiving and extracting from the received bitstream, for example a demultiplexer, the quantized filter LPC4; and

[0097]
an absolute inverse quantizer supplied with the quantized filter LPC4 for inverse quantizing the quantized filter LPC4.

[0098]
Operation 304: If the superframe is coded using mode TCX1024, no further transmission is required.

[0099]
Operation 305: If the first, second, third and fourth frames of the superframe are not coded using mode TCX1024, the quantized filter LPC2 and an index indicative of one of the absolute quantization mode and the differential quantization mode are transmitted to the decoder for example through the transmitter 105 of FIG. 1. The decoder comprises:

[0100]
means for receiving and extracting from the received bitstream, for example a demultiplexer, the quantized filter LPC2 and the index indicative of one of the absolute quantization mode and the differential quantization mode; and

[0101]
an absolute inverse quantizer supplied with the quantized filter LPC2 and the index indicative of the absolute quantization mode for inverse quantizing the quantized filter LPC2, or a differential inverse quantizer supplied with the quantized filter LPC2 and the index indicative of the differential quantization mode for inverse quantizing the quantized filter LPC2.

[0102]
Operation 306: If frames 1 and 2 of the superframe are coded using mode TCX512, the quantized filter LPC1 is not transmitted to the decoder.

[0103]
Operation 307: If frames 1 and 2 of the superframe are not coded using mode TCX512, i.e. if frames 1 and 2 of the superframe are coded using ACELP or TCX256, the quantized filter LPC1, and an index indicative of one of the absolute quantization mode, the differential quantization mode relative to quantized filter LPC2 used as reference, and the differential quantization mode relative to (quantized filter LPC2+quantized filter LPC0)/2 used as reference are transmitted to the decoder for example through the transmitter 105 of FIG. 1. The decoder comprises:

[0104]
means for receiving and extracting from the received bitstream, for example a demultiplexer, the quantized filter LPC1, and the index indicative of one of the absolute quantization mode, the differential quantization mode relative to quantized filter LPC2 used as reference, and the differential quantization mode relative to (quantized filter LPC2+quantized filter LPC0)/2 used as reference; and

[0105]
an absolute inverse quantizer supplied with the quantized filter LPC1 and the index indicative of one of the absolute quantization mode, the differential quantization mode relative to quantized filter LPC2 used as reference, and the differential quantization mode relative to (quantized filter LPC2+quantized filter LPC0)/2) used as reference for inverse quantizing the quantized filter LPC1.

[0106]
Operation 308: If frames 3 and 4 of the superframe are coded using mode TCX512, the quantized filter LPC3 is not transmitted to the decoder.

[0107]
Operation 309: If frames 3 and 4 of the superframe are not coded using mode TCX512, i.e. if frames 3 and 4 of the superframe are coded using ACELP or TCX256, the quantized filter LPC3 and the index indicative of one of the absolute quantization mode, the differential quantization mode relative to quantized filter LPC2 used as reference, the differential quantization mode relative to quantized filter LPC4 used as reference, and the differential quantization mode relative to (quantized filter LPC2+quantized filter LPC4)/2 used as reference are transmitted to the decoder for example through the transmitter 105 of FIG. 1. The decoder comprises:

[0108]
means for receiving and extracting from the received bitstream, for example a demultiplexer, the quantized filter LPC3, and the index indicative of one of the absolute quantization mode, the differential quantization mode relative to quantized filter LPC2 used as reference, the differential quantization mode relative to quantized filter LPC4 used as reference, and the differential quantization mode relative to (quantized filter LPC2+quantized filter LPC4)/2 used as reference; and

[0109]
an absolute inverse quantizer supplied with the quantized filter LPC3 and the index indicative of one of the absolute quantization mode, the differential quantization mode relative to quantized filter LPC2 used as reference, the differential quantization mode relative to quantized filter LPC4 used as reference, and the differential quantization mode relative to (quantized filter LPC2+quantized filter LPC4)/2 used as reference for inverse quantizing the quantized filter LPC3.

[0110]
Some benefits of the above described solution comprise:

 Quantizing the whole set of LPC filters prior to the closedloop selection of the coding modes saves complexity;
 Using a differential quantizer in the global quantization scheme preserves some of the bit rate saving that was gained by, for example, the predictive quantizer in the original AMRWB+ quantization scheme.

[0113]
The following variants can be used to build the reference LPC filters that are used in the differential quantizers (Operations such as
103 _{1},
103 _{2}, . . . ,
103 _{n}):

 If inter superframe dependency is not an issue, the last LPC filter (LPC4) from the previous superframe (LPC0) can be used as a possible reference for encoding the filters LPC1 to LPC4;
 When different reference LPC filters are available, for example filter LPC0 and LPC4 when coding filter LPC2, a specific bit pattern can be transmitted to the decoder to indicate which of the references is actually used. For example, selection of the reference can be performed as described hereinabove with reference to FIG. 1, for example on the basis of a distance or a bit rate measurement.
 When different reference LPC filters are available, additional secondary reference LPC filters can be obtained by applying various extrapolations or interpolations schemes to the already available reference LPC filters. A specific bit pattern can be transmitted to indicate the actual interpolation or extrapolation strategy selected by the coder. For example, filter LPC3 can be quantized differentially with respect to the quantized versions of either filter LPC2 or LPC4, or even with respect to an interpolated value (e.g. average) between these two quantized filters LPC2 and LPC4 (see Operation 505 of FIG. 5).

[0117]
The above described “outofthe loop” quantization scheme can be extended to coding more than four (4) LPC filters: for example to quantize and transmit filter LPC0 together with the superframe. In that case, filter LPC0 corresponding to the last LCP filter (LPC4) calculated during the previous superframe could be, as a nonlimitative example, be quantized relative to filter LPC4 since this filter LPC4 is always available as a reference. Quantized filter LPC0 is transmitted to the decoder along with an index indicative of one of the absolute quantization mode and the differential quantization mode. The decoder comprises:

[0118]
means for receiving and extracting from the received bitstream, for example a demultiplexer, the quantized filter LPC0, and the index indicative of one of the absolute quantization mode and the differential quantization mode; and

[0119]
an absolute inverse quantizer supplied with the quantized filter LPC0 and the index indicative of the absolute quantization mode for inverse quantizing the quantized filter LPC0, or a differential inverse quantizer supplied with the quantized filter LPC0, and the index indicative of the differential quantization mode for inverse quantizing the quantized filter LPC0.

[0120]
Transmitting filter LPC0 to the decoder is useful to initialize an LPCbased codec in the case of switching from a nonLPCbased coding mode to an LPCbased coding mode. Examples of nonLPCbased coding modes are: pulse code modulation (PCM), and transform coding used for example by MP3 and by the advanced audio codec AAC. Examples of LPCbased coding modes are: code excited linear prediction (CELP) and algebraic CELP (ACELP) used by the AMRWB+ codec [1].

[0121]
In LPCbased codecs, one or several LPC filters per frame (or per superframe) are estimated and transmitted to the decoder. When one single LPC filter per frame is estimated and transmitted, this LPC filter is most often estimated using an LPC analysis window centered on the end of the frame as represented in FIG. 4 a. When several LPC filters are transmitted per frame (or per superframe as in the AMRWB+codec), they are most often estimated at regularly spaced positions over the length of the frame as represented on FIG. 4 b. Filter LPC0 in FIGS. 4 a and 4 b is in fact the last LPC filter of the previous frame (or superframe) which is quantized and transmitted to the decoder.

[0122]
Typical LPCbased codecs generally use interpolated values for the LPC filters. In the example of FIG. 4 a, for example, the LPCbased codec would typically divide the frame into four (4) subframes and use a different interpolated LPC filter for each subframe, the LPC filter of the first subframe being closer to filter LPC0 and the LPC filter of the 4^{th }subframe being closer to filter LPC1.

[0123]
In a codec which switches from a nonLPCbased coding mode to an LPCbased coding mode, the filter LPC0 used to operate the LPCbased codec is normally not available at the first frame following the switch from the nonLPCbased coding mode to the LPCbased coding mode.

[0124]
In that context, it is proposed to provide a value for filter LPC0 which is available both at the coder and the decoder when coding and decoding the first frame after the switch from the nonLPCbased coding mode to the LPCbased coding mode. More specifically, the value of the filter LPC0 is obtained at the decoder from the parameters transmitted from the coder.

[0125]
According to a first solution, the filter LPC0 is determined at the coder (using LPC analysis wellknow to those of ordinary skill in the art), quantized and transmitted to the decoder after the switch from the nonLPCbased coding mode to the LPCbased coding mode has been decided. The decoder uses the transmitted quantized value and the filter LPC0. To quantize the filter LPC0 efficiently, the outoftheloop quantization scheme as described above, extended to more than four (4) LPC filters can be used.

[0126]
The following describes second and third solutions to estimate the filter LPC
0 at the decoder from transmitted parameters:

 Estimation of the filter LPC0 from the other transmitted LPC filters using, for example, extrapolation; and
 Estimation of the filter LPC0 from the other transmitted parameters. For example the filter LPC0 can be estimated by applying the conventional LPC analysis procedure to the past decoded signal, more specifically the output of the switched decoder prior to the switch from the nonLPCbased coding mode to the LPCbased coding mode.
Quantization with a Uniform Algebraic Vector Quantizer

[0129]
The principle of stochastic vector quantization is to search a codebook of vectors for the nearest neighbor (generally in terms of Euclidian distance or weighted Euclidian distance) of the vector to be quantized. When quantizing LPC filters in the LSF (Line Spectral Frequency) or ISF (Immitance Spectral Frequency) domains, a weighting Euclidian distance is generally used, each component of the vector being weighted differently depending on its value and the value of the other components [5]. The purpose of that weighting is to make the minimization of the Euclidian distance behave as closely as possible to a minimization of the spectral distortion. Unlike a stochastic quantizer, a uniform algebraic vector quantizer does not perform an exhaustive search of a codebook. It is therefore difficult to introduce a weighting function in the distance computation.

[0130]
In the solution proposed herein, the LPC filters are quantized, as a nonlimitative example, in the LSF domain. Appropriate means for converting the LPC filter in the LSF quantization domain to form the input LSF vector are therefore provided. More specifically, the LSF residual vector, i.e. the difference between the input LSF vector and a firststage approximation of this input LSF vector, is warped using a weighting function computed from the firststage approximation, wherein the firststage approximation uses a stochastic absolute quantizer of the input LSF vector, a differential quantizer of the input LSF vector, an interpolator of the input LSF vector, or other element that gives an estimate of the input LSF vector to be quantized. Warping means that different weights are applied to the components of the LSF residual vector. Because the firststage approximation is also available at the decoder, the inverse weights can also be computed at the decoder and the inverse warping can be applied to the quantized LSF residual vector. Warping the LSF residual vector according to a model that minimizes the spectral distortion is useful when the quantizer is uniform. The quantized LSFs received at the decoder are a combination of the firststage approximation with a variable bit rate quantization, for example AVQ (Algebraic Vector Quantization), refinement which is inversewarped at the decoder.

[0131]
Some benefits of the proposed solution are the following:

 With a good weighting function, a uniform quantizer can provide a relatively uniform spectral distortion.
 The advantages of variable bit rate vector quantization, for example AVQ (Algebraic Vector Quantization), over SVQ (Stochastic Vector Quantization) are a smaller number of tables (memory), less complexity and higher bitrate granularity.
 Another advantage in favor of variable bit rate vector quantization, for example AVQ (Algebraic Vector Quantization), is its unlimited codebook size; this guarantees the same spectral distortion for any type of signal.

[0135]
The general principle for the quantization of a given LPC filter is given in FIG. 6. In this nonlimitative example, the LPC filter is quantized in the LSF domain.

[0136]
Operation 601: A calculator computes a firststage approximation 608 of the input LSF vector 607.

[0137]
Operation 602: A subtractor subtracts the firststage approximation 608 from Operation 601 from the input LSF vector 607 to produce a residual LSF vector 609.

[0138]
Operation 603: A calculator derives a LSF weighting function 610 from the firststage approximation 608 of Operation 601.

[0139]
Operation 604: A multiplier, or warper, applies the LSF weighting function 610 from Operation 603 to the residual LSF vector 609 from Operation 602.

[0140]
Operation 605: A variable bit rate quantizer, for example an algebraic vector quantizer (AVQ) quantizes the resulting weighted residual LSF vector 611 to supply a quantized weighted residual LSF vector 612.

[0141]
Operation 606: A multiplexer is responsive to the firststage approximation 608 from Operation 601 and the quantized weighted residual LSF vector 612 from Operation 605 to multiplex and transmit the corresponding coded indices 613.

[0142]
The firststage approximation (Operation 601) can be calculated in different ways. As a nonlimitative example, the calculator of the firststage approximation 608 can be an absolute stochastic vector quantizer of the input LSF vector 607 with a small number of bits, or a differential quantizer of the input LSF vector 607 using a reference as explained above where the firststage approximation is the reference itself. For example, when quantizing the vector LPC1 as in FIG. 5, Operation 504, the calculator of the firststage approximation 608 can be an absolute quantizer with 8 bits, or quantized filter LPC2 or (quantized filter LPC2+quantized filter LPC0)/2.

[0143]
The calculation and purpose of the weighting function (Operation 603) is described herein below.

[0144]
The corresponding inverse quantizer is illustrated in FIG. 7.

[0145]
Operation 701: The coded indices 707 from the coder are demultiplexed by a demultiplexer.

[0146]
Operation 702: The demultiplexed coded indices include the firststage approximation 708.

[0147]
Operation 703: Since the firststage approximation is available at the decoder as at the coder (Operation 702), a calculator can be used to calculate the inverse LSF weighting function 709.

[0148]
Operation 704: Decoded indices 710 representative of the quantized weighted residual LSF vector are supplied to a variable bit rate inverse vector quantizer, for example an algebraic inverse vector quantizer (inverse AVQ) to recover the weighted residual LSF vector 711.

[0149]
Operation 705: A multiplier multiplies the weighted residual LSF vector 711 from Operation 704 by the inverse LSF weighting function 709 from Operation 703 to recover the residual LSF vector 712.

[0150]
Operation 706: An adder sums the firststage approximation 708 from Operation 702 with the residual LSF vector 712 from Operation 705 to form the decoded LSF vector 713. The decoded LSF vector 713 is a combination of the firststage approximation from Operation 702 with the variable bit rate inverse quantization refinement (Operation 704) which is inverseweighted (Operation 705) at the decoder.
FirstStage Approximation

[0151]
As explained above a given LPC filter can be quantized using several quantization modes including absolute quantization and differential quantization using several references. The firststage approximation depends on the quantization mode. In the case of absolute quantization, the firststage approximation can use a vector quantizer with a small number of bits (e.g. 8 bits). In the case of differential quantization, the firststage approximation constitutes the reference itself. For example, when quantizing the vector LPC
3 as illustrated in
FIG. 5 (Operation
505), the firststage approximation can be one of the following:

 8bit VQ (absolute quantization);
 Quantized filter LPC2 (differential quantization using quantized filter LPC2 as reference);
 Quantized filter LPC4 (differential quantization using quantized filter LPC4 as reference); or
 Average of quantized filters LPC2 and LPC4 (differential quantization using (quantized filter LPC2+quantized filter LPC4)/2 as reference).

[0156]
As a nonlimitative example, in the case of a pth order LPC filter expressed with LSF parameters, in the absolute quantization mode, the firststage approximation is calculated using a pdimensional, 8bit stochastic vector quantizer applied to the input LSF vector. A codebook search uses a weighted Euclidian distance in which each component of the squared difference between the input LSF vector and the codebook entry is multiplied by the weight wt(i). For example, the weight wt(i) can be given by the following expression:

[0000]
$\mathrm{wt}\ue8a0\left(i\right)=\frac{1}{{d}_{i}}+\frac{1}{{d}_{i+1}},\text{}\ue89ei=0,\dots \ue89e\phantom{\rule{0.8em}{0.8ex}},p1$

[0000]
with:

[0000]
d _{0} =f(0)

[0000]
d _{p} =SF/2−f(p−1)

[0000]
d _{i} =f(i)−f(i−1),i=1, . . . , p−1

[0000]
where f(i), i=0, . . . , p−1 is the input LSF vector to be quantized, p is the order of LP analysis, and SF is the internal sampling frequency of the LPCbased codec (in Hz).

[0157]
In the differential quantization modes, the firststage approximation is based on already quantized LPC filters.

[0158]
As explained with reference to FIG. 5, the set of LPC filters is quantized in the following order: LPC4, LPC2, LPC1 and then LPC3. When required, the optional filter LPC0 is quantized after the filter LPC4. Therefore differential quantization of filter LPC2 can only be done with respect to LPC4, while differential quantization of filter LPC3 can be done with respect to LPC2, LPC4 or a combination of both LPC2 and LPC4; LPC1 is not considered a good choice because it is not adjacent to LPC3.

[0159]
For each firststage approximation f_{1st}(i), the residual LSF vector is calculated as:

[0000]
r(i)=f(i)−f _{1st}(i),i=0, . . . , p−1

[0160]
As shown in FIG. 6, the residual LSF vector 609 from Operation 602 is weighted (Operation 604) with the weighting function 610 from Operation 603 computed based on the firststage approximation f_{1st}(i) to obtain a warped residual LSF vector 611 (Operation 604). The warped residual LSF vector 611 is then quantized using a variable bit rate quantizer, for example an algebraic vector quantizer (Operation 605).

[0161]
For example, the weights applied to the components of the pth residual LSF vector can be given by the following relation:

[0000]
$w\ue8a0\left(i\right)=\frac{1}{W}\star \frac{400}{\sqrt{{d}_{i}\xb7{d}_{i+1}}},\text{}\ue89ei=0,\dots \ue89e\phantom{\rule{0.8em}{0.8ex}},p1$

[0000]
with:

[0000]
d _{0} =f _{1st}(0)

[0000]
d _{p} =SF/2−f _{1st}(p−1)

[0000]
where f_{1st}(i) is the firststage approximation, SF is the internal sampling frequency in Hz of the LPCbased codec, and W is a scaling factor which depends on the quantization mode. The value of W is chosen in order to obtain a certain target spectral distortion and/or a certain target average bit rate once the warped residual LSF vector is quantized with the variable bit rate quantizer. As a nonlimitative example, the variable bit rate vector quantizer chooses the bit rate for a certain vector based on its average energy.

[0162]
In an illustrative example, the four (4) LPC filters in a superframe, as well as the optional LPC0 filter are quantized according to FIG. 5. Table 2 shows the used scaling factor for each quantization mode, and the encoding of the mode index used in this example. Note that the quantization mode specifies which of the absolute or differential quantization is used, and in case of differential quantization it specifies the used reference filter. As explained above the reference filter used in differential quantization is the actual firststage approximation for variable bit rate quantizing.

[0000]
TABLE 2 

Possible absolute and relative quantization modes 
and corresponding bitstream signaling, and the 
scaling factor and the weighting function 

Quantization 
First stage 
Encoded 

Filter 
mode 
approximation 
mode 
W 

LPC4 
Absolute 
8bit VQ 
(none) 
60 
LPC0 
Absolute 
8bit VQ 
0 
60 

Relative LPC4 
Quantized LPC4 
1 
63 
LPC2 
Absolute 
8bit VQ 
0 
60 

Relative LPC4 
Quantized LPC4 
1 
63 
LPC1 
Absolute 
8bit VQ 
00 
60 

Relative 
Quantized 
01 
65 

(LPC0 + LPC2)/2 
(LPC0 + LPC2)/2 

(Note 1) 

Relative LPC2 
Quantized LPC2 
10 
64 
LPC3 
Absolute 
8bit VQ 
10 
60 

Relative 
Quantized 
0 
65 

(LPC2 + LPC4)/2 
(LPC2 + LPC4)/2 

Relative LPC2 
Quantized LPC2 
110 
64 

Relative LPC4 
Quantized LPC4 
111 
64 

(Note 1): 
in this mode, there is no secondstage AVQ quantizer 

[0163]
FIG. 8 is a schematic block diagram explaining the quantization procedure as described herein above.

[0164]
Operations 801, 801 _{1}, 801 _{2}, . . . , 801 _{n}: The input LSF vector 800 is supplied to an absolute quantizer (Operation 801) for performing, for example, a 8bit absolute vector quantization of the input LSF vector 800. The input LSF vector is also supplied to differential quantizers (Operations 801 _{1}, 801 _{2}, . . . , 801 _{n}) for performing differential quantization of the input LSF vector 800. The differential quantizers use respective, different references as explained in the foregoing description with reference to FIG. 1. The 8bit VQ in Operation 801 and references in Operations 801 _{1}, 801 _{2}, 801 _{n }represent the firststage approximation.

[0165]
In Operations 802, 802 _{1}, 802 _{2}, . . . , 802 _{n}, a calculator calculates a residual LSF vector from the first stage approximation vector from the Operations 801, 801 _{1}, 801 _{2}, . . . , 801 _{n}, respectively. The residual vector is calculated as the difference between the input vector and the firststage approximation. This corresponds to Operations 601 and 602 of FIG. 6.

[0166]
In Operations 803, 803 _{1}, 803 _{2}, . . . , 803 _{n}, a calculator calculates a weighting function to warp the residual LSF vector from the Operations 802, 802 _{1}, 802 _{2}, . . . , 802 _{n}, respectively. This corresponds to Operations 601 and 603 of FIG. 6.

[0167]
In Operations 804, 804 _{1}, 804 _{2}, . . . , 804 _{n}, a warper multiplies the residual LSF vector from the Operations 802, 802 _{1}, 802 _{2}, . . . , 802 _{n}, respectively, by the weighting function from the Operations 803, 803 _{1}, 803 _{2}, . . . 803 _{n}, respectively.

[0168]
In Operations 805, 805 _{1}, 805 _{2}, . . . , 805 _{n }a variable bit rate quantizer, for example an algebraic vector quantizer (AVQ) quantizes the resulting weighted residual LSF vector from the Operations 804, 804 _{1}, 804 _{2}, . . . , 804 _{n}, respectively, to supply a quantized weighted residual LSF vector.

[0169]
In Operation 806, the selection of a quantization mode is performed by a selector amongst absolute quantization (Operation 801) and differential quantization using one of the references 1, 2, . . . , n (Operations 801 _{1}, 801 _{2}, . . . , 801 _{n}). For example, Operation 806 could select the quantization mode (Operations 801, 801 _{1}, 801 _{2}, . . . , 801 _{n}) that yields a lower distortion for a given bit rate or the lower bit rate for a target level of distortion. Regarding the selection amongst 8bit VQ and references 1, 2, . . . , n, the selection can be performed in closedloop or in openloop. In closedloop, all possible references are tried and the reference that optimizes a certain criterion of distortion or bit rate is chosen, for example the lower distortion for a given bit rate or the Lower bit rate for a target level of distortion. In openloop, the Operation 806 predetermines the reference based on the value of the Linear Prediction Coefficients of the LPC filter to be quantized and of the Linear Prediction Coefficients of the available reference LPC filters.

[0170]
Operation
807: Following the selection in Operation
806, a transmitter (Operation
807) communicates or signals to the decoder (not shown) an index indicative of:

 the quantization mode (suboperation 807 _{1}), for example absolute or differential quantization; and

[0172]
in the case of differential quantization, of the selected reference and associated differential quantizer of Operations 801 _{1}, 801 _{2}, . . . , 801 _{n }(suboperation 807 _{2}).

[0173]
Some specific bits are transmitted to the decoder for such signaling.
Algebraic Vector Quantizer

[0174]
A possible algebraic vector quantizer (AVQ) used for example in Operation 605 of FIG. 6 and Operations 805, 805 _{1}, 805 _{2}, . . . , 805 _{n }of FIG. 8 is based on the 8dimensional RE8 lattice vector quantizer used to quantize the spectrum in TCX modes of AMRWB+[1].

[0175]
For a 16^{th }order LPC, each weighted residual LSF vector is split into two 8dimensional subvectors B_{1 }and B_{2}. Each of these two subvectors is quantized using the threeoperation approach described below.

[0176]
LSF vectors do not have all the same sensitivity to quantization error, whereby a certain quantization error applied to one LSF vector can have more impact on spectral distortion than the same quantization error applied to another LSF vector. The weighting operation gives the same relative sensitivity to all weighted LSF vectors. The AVQ has the particularity of introducing the same level of quantization error to the weighted LSF vectors (uniform quantization error). When performing the inverse quantization, the inverse weighting which is applied to inversequantized weighted LSF vectors is also obviously applied to the quantization error. Thus, the originally uniform quantization error is distributed among quantized LSF vectors, the more sensitive LSF vectors acquiring a smaller quantization error and the less sensitive LSF vectors acquiring a larger quantization error. As a consequence, the impact of quantization error on spectral distortion is minimized.

[0177]
As explained in Reference [1], the RE8 quantizer uses a fixed and predetermined quantization. As a consequence, the bit rate required to encode a subvector increases with the amplitude of this subvector.

[0178]
The scaling factor W controls the amplitude of the weighted LSF vectors. Therefore, the scaling factor W also controls both the bit rate needed to quantize the LSF vector and the average spectral distortion.
First Operation: Find Nearest Neighbor in Lattice RE_{8 }

[0179]
In this first operation, an 8dimensional subvector B_{k }is rounded as a point in the lattice RE_{8}, to produce its quantized version, {circumflex over (B)}_{k}. Before looking at the quantization procedure, it is worthwhile to look at the properties of this lattice. The lattice RE_{8 }is defined as follows:

[0000]
RE _{8}=2D _{8}∪{2D _{8}+(1, . . . , 1)}

[0000]
that is as the union of a lattice 2D_{8 }and a version of the lattice 2D_{8 }shifted by the vector (1, 1, 1, 1, 1, 1, 1, 1). Therefore, searching for the nearest neighbour in the lattice RE_{8 }is equivalent to searching for the nearest neighbour in the lattice 2D_{8}, then searching for the nearest neighbour in the lattice 2D_{8}+(1, 1, 1, 1, 1, 1, 1, 1), and finally selecting the best of those two lattice points. The lattice 2D_{8 }is the lattice D_{8 }scaled by a factor of 2, with the lattice D_{8 }defined as:

[0000]
D _{8}={(x _{1} , . . . , x _{8})εZ ^{8} x _{1} + . . . +x _{8 }is even}

[0000]
That is, the points in the lattice D_{8 }are all integers, with the constraint that the sum of all components is even. This also implies that the sum of the components of a point in lattice 2D_{8 }is an integer multiple of 4.

[0180]
From this definition of lattice RE
_{8}, it is straightforward to develop a fast algorithm to search for the nearest neighbour of an 8dimensional subvector B
_{k }among all lattice points in lattice RE
_{8}. This can be done by applying the following operations. The components of subvector B
_{k }are floating point values, and the result of the quantization, {circumflex over (B)}
_{k}, will be a vector of integers.

 1. z_{k}=0.5*B_{k }
 2. Round each component of z_{k }to the nearest integer, to generate z _{k }
 3. y1_{k}=2 z _{k }
 4. Calculate S as the sum of the components of y1_{k }
 5. If S is not an integer multiple of 4 (negative values are possible), then modify one of its components as follows:
 find the position I where abs(z_{k}(i)−y1_{k}(i)) is the highest

[0000]
if z _{k}(I)−y1_{k}(I)<0, then y1_{k}(I)=y1_{k}(I)−2

[0000]
if
z _{k}(
I)−
y1
_{k}(
I)>0, then
y1
_{k}(
I)=
y1
_{k}(
I)+2

 6. z_{k}=0.5*(B_{k}−1.0) where 1.0 denotes a vector in which all the components are 1's
 7. Round each component of z_{k }to the nearest integer, to generate z _{k }
 8. y2_{k}=2 z _{k }
 9. Calculate S as the sum of the components of y2_{k }
 10. If S is not an integer multiple of 4 (negative values are possible), then modify one of its components as follows:
 find the position I where abs(z_{k}(I)−y2_{k}(I)) is the highest

[0000]
if z _{k}(I)−y2_{k}(I)<0, then y2_{k}(I)=y2_{k}(I)−2

[0000]
if
z _{k}(
I)−
y2
_{k}(
I)>0, then
y2
_{k}(
I)=
y2
_{k}(
I)+2

 11. y2_{k}=y2_{k}+1.0
 12. Compute e1 _{k}=(B_{k}−y1_{k})^{2 }and e2_{k}=(B_{k}−y2_{k})^{2 }
 13. If e1_{k}>e2_{k}, then the best lattice point (nearest neighbour in the lattice) is y1_{k }otherwise the best lattice point is y2_{k}.
{circumflex over (B)}_{k}=c_{k }where c_{k }is the best lattice point as selected above.
Second Operation: Computation of the Indices

[0196]
In the first operation, each 8dimensional subvector B_{k }was rounded as a point in the lattice RE_{8}. The result is {circumflex over (B)}_{k}=c_{k}, the quantized version of B_{k}. In the present second operation an index is computed for each c_{k }for transmission to the decoder. The computation of this index is performed as follows.

[0197]
The calculation of an index for a given point in the lattice RE
_{8 }is based on two basic principles:
 1. All points in the lattice RE_{8 }lie on concentric spheres of radius √{square root over (8m)} with m=0, 1, 2, 3, etc., and each lattice point on a given sphere can be generated by permuting the coordinates of reference points called leaders. There are very few leaders on a sphere, compared to the total number of lattice points which lie on the sphere. Codebooks of different bit rates can be constructed by including only spheres up to a given number m. See Reference [6] for more details, where codebooks Q_{0}, Q_{1}, Q_{2}, Q_{3}, Q_{4}, and Q_{5 }are constructed with respectively 0, 4, 8, 12, 16 and 20 bits. Hence, codebook Q_{n }requires 4n bits to index any point in that codebook.
 2. From a base codebook C (i.e. a codebook containing all lattice points from a given set of spheres up to a number m), an extended codebook can be generated by multiplying the elements of the base codebook C by a factor M, and adding a secondstage codebook called the Voronoi extension. This construction is given by y=Mz+v, where M is the scale factor, z is a point in the base codebook and v is the Voronoi extension. The extension is computed in such a way that any point y=Mz+v is also a point in the lattice RE_{8}. The extended codebook includes lattice points that extend further out from the origin than the base codebook.

[0200]
In the present case, the base codebook C in the LPC quantizer can be either codebook Q_{0}, Q_{2}, Q_{3 }or Q_{4 }from Reference [6]. When a given lattice point c_{k }is not included in these base codebooks, the Voronoi extension is applied, using this time only the codebook Q_{3 }or Q_{4}. Note that here, Q_{2}⊂Q_{3 }but Q_{3}⊂Q_{4}.

[0201]
Then, the calculation of the index for each lattice point c
_{k}, obtained in the first operation, is performed according to the following operations.

 Verify if c_{k }is in the base codebook C. This implies verifying if c_{k }is an element of codebooks Q_{0}, Q_{2}, Q_{3 }or Q_{4 }from Reference [6].
 If c_{k }is an element of the base codebook C, the index used to encode c_{k }is thus the codebook number n_{k }plus the index I_{k }of the codevector c_{k }in the codebook Q_{n} _{ k }. The codebook number n_{k }is encoded as described in a third operation. The index I_{k }indicates the rank of the codevector c_{k}, i.e. the permutation to be applied to a specific leader to obtain c_{k }(see Reference [7]). If n_{k}=0, then I_{k }uses no bits. Otherwise, the index I_{k }uses 4n_{k }bits.
 If c_{k }is not in the base codebook, then apply the Voronoi extension through the following suboperations, using this time only the codebook Q_{3 }or Q_{4 }as the base codebook.
 V0 Set the extension order r=1 and the scale factor M=2^{r}=2.
 V1 Compute the Voronoi index k of the lattice point c_{k}. The Voronoi index k depends on the extension order r and the scale factor M. The Voronoi index is computed via modulo operations such that k depends only on the relative position of c_{k }in a scaled and translated Voronoi region:

[0000]
k=mod
_{M}(
c _{k} G ^{−1}).



 where G is the generator matrix and mod_{M}(•) is the componentwise moduloM operation. Hence, the Voronoi index k is a vector of integers with each component comprised in the interval 0 to M−1.
 V2 Compute the Voronoi codevector v from the Voronoi index k. This can be implemented using an algorithm as described in reference [8].
 V3 Compute the difference vector w=c_{k}−v. This difference vector w always belongs to the scaled lattice mΛ, where Λ is the lattice RE_{8}. Compute z=w/M, i.e., apply the inverse scaling to the difference vector w. The codevector z belongs to the lattice Λ since w belongs to MΛ.
 V4 Verify if z is in the base codebook C (i.e. in Q_{3 }or Q_{4})
 If z is not in the base codebook C, increment the extension order r by 1, multiply the scale factor M by 2, and go back to suboperation V1. Otherwise, if z is in the base codebook C, then an extension order r and a scaling factor M=2^{r }sufficiently large to encode the index of c_{k }has been found. The index is formed of three parts: 1) the codebook number n_{k }as a unary code defined below; 2) the rank I_{k }of z in the corresponding base codebook (either Q_{3 }or Q_{4}); and 3) the 8 indices of the Voronoi index vector k calculated in suboperation V1, where each index requires exactly r bits (r is the Voronoi extension order set in suboperation V0). The codebook number n_{k }is encoded as described in the third operation.

[0212]
The lattice point c_{k }is then described as:

[0000]
c
_{k}
=Mz+v.
Third Operation: Variable Length Coding of the Codebook Numbers

[0213]
The codebook numbers n_{k }are encoded using a variablelength code which depends on the position of the LPC filter and on the quantization mode, as indicated in Table 3.

[0000]
TABLE 3 

Coding modes for codebook numbers n_{k} 
Filter  Quantization mode  n_{k }mode 

LPC4  Absolute  0 
LPC0  Absolute  0 
 Relative LPC4  3 
LPC2  Absolute  0 
 Relative LPC4  3 
LPC1  Absolute  0 
 Relative (LPC0 + LPC2)/2  1 
 Relative LPC2  2 
LPC3  Absolute  0 
 Relative (LPC2 + LPC4)/2  1 
 Relative LPC2  2 
 Relative LPC4  2 

n
_{k }modes 0 and 3:

[0214]
The codebook number n
_{k }is encoded as a variable length code, as follows:

 Q_{2}→the code for n_{k }is 00
 Q_{3}→the code for n_{k }is 01
 Q_{4}→the code for n_{k }is 10

[0218]
Others: the code for n
_{k }is 11 followed by:

 Q_{5}→0
 Q_{6}→10
 Q_{7}→1110
 Q_{8}→11110
 etc.
n_{k }Mode 1:

[0224]
The codebook number n
_{k }is encoded as a unary code, as follows:

 Q_{0}→unary code for n_{k }is 0
 Q_{2}→unary code for n_{k }is 10
 Q_{3}→unary code for n_{k }is 110
 Q_{4}→unary code for n_{k }is 1110
 etc.
n_{k }Mode 2:

[0230]
The codebook number n
_{k }is encoded as a variable length code, as follows:

 Q_{2}→the code for n_{k }is 00
 Q_{3}→the code for n_{k }is 01
 Q_{4}→the code for n_{k }is 10

[0234]
Others: the code for n
_{k }is 11 followed by:

 Q_{0}→0
 Q_{5}→10
 Q_{6}→110
 etc.
Quantization Mode Decision

[0239]
For each LSF vector, all possible absolute and differential quantization modes as described in Table 2 are each tried and, for example, the quantization mode which requires the minimum number of bits is selected. The encoded quantization mode and the corresponding set of quantization indices are transmitted to the decoder.

[0240]
As mentioned in the foregoing description, the actual number of quantized LPC filters transmitted from the coder to the decoder is not fixed but rather depends on the ACELP/TCX decision taken at the coder. For example, long TCX (TCX1024) requires only the transmission of quantized filter LPC4 while any combination involving ACELP or short TCX (TCX256) requires the transmission of all four (4) quantized LPC filters LPC1 to LPC4. Only the quantized LPC filters that are required by the ACELP/TCX mode configuration are actually transmitted.
Decoding Process of Algebraic Vector Quantizer

[0241]
As mentioned herein above, the actual number of quantized LPC filters coded within the bitstream depends on the ACELP/TCX mode combination of the superframe. The ACELP/TCX mode combination is extracted from the bitstream and determines the coding modes, mod [k] for k=0 to 3, of each of the four (4) frames composing the superframe. The mode value is 0 for ACELP, 1 for TCX256, 2 for TCX512, 3 for TCX1024.

[0242]
In addition to the one (1) to four (4) quantized LPC filters of the superframe, the above described, optional quantized filter LPC0 is transmitted for the first superframe of each segment coded using the linearprediction based codec.

[0243]
The order in which the quantized LPC filters are normally found in the bitstream is: LPC4, the optional LPC0, LPC2, LPC1, and LPC3.

[0244]
The condition for the presence of a given LPC filter within the bitstream is summarized in Table 4.

[0000]
TABLE 4 

Condition for the presence of a given LPC filter in the bitstream 

LPC filter 
Present if 



LPC0 
1st superframe encoded using LP 

LPC1 
mod[0] < 2 

LPC2 
mod[2] < 3 

LPC3 
mod[2] < 2 

LPC4 
Always 



[0245]
FIG. 9 is a schematic block diagram summarizing the decoding process.

[0246]
Operations 901 and 902: The decoder comprises means for receiving and extracting, for example a demultiplexer, from the received bit stream the quantization indices corresponding to each of the quantized LPC filters required by the ACELP/TCX mode combination. For a given quantized LPC filter, a determiner of quantization mode extracts from the bit stream received from the coder the index or information related to the quantization mode, and determines whether the quantization mode is the absolute or differential quantization mode as indicated in Table 2.

[0247]
Operations 903 and 905: When Operations 901 and 902 determine that the quantization mode is the absolute quantization mode, an extractor extracts from the bit stream the index or indices corresponding to the stochastic VQquantized firststage approximation (Operation 903). A calculator then computes the firststage approximation through inversequantization (Operation 905).

[0248]
Operations 904 and 905: When Operations 901 and 902 determine that the quantization mode is the differential quantization mode (not the absolute quantization mode), an extractor extracts from the bit stream the indices or information representative of the reference amongst the plurality of possible references, for example the reference LPC vector (Operation 904). The calculator then computes from this information the firststage approximation as described with reference to Table 2 (Operation 905).

[0249]
In Operation 906, an extractor of VQ information extracts from the bit stream received from the coder variable bit rate VQ information, for example AVQ information. More specifically, as a nonlimitative example, the AVQ information for the two residual LSF subvectors {circumflex over (B)}_{k }are extracted from the bit stream. The AVQ information normally comprises two encoded codebook numbers and the corresponding AVQ indices. The only exception is when filter LPC1 is differentially quantized relative to (quantized filter LPC0+quantized filter LPC2)/2, since in this case there is no AVQ information present in the bit stream. In the case of the latter exception, the quantized LSF vector 909 is outputted as the firststage approximation from Operation 905.

[0250]
Operation 907: An inverse algebraic vector quantizer receives the extracted AVQ information from Operation 906 to inverse quantize, or inverse weight and recover the AVQ contribution.
Decoding of AVQ Indices

[0251]
Decoding the LPC filters involves decoding the extracted AVQ information, for example the AVQ parameters describing each quantized subvector {circumflex over (B)}
_{k }of the weighted residual LSF vector. In the foregoing example, each subvector B
_{k }has a dimension
8. The AVQ parameters for each subvector B
_{k }are described in the second operation of the above described algebraic vector quantization. For each quantized subvector {circumflex over (B)}
_{k}, three sets of binary indices are sent by the coder to the decoder:

 a) the codebook number n_{k}, transmitted using an entropy code as described in the third operation of the above described algebraic vector quantization;
 b) the rank I_{k }of a selected lattice point z in a base codebook, which indicates what permutation has to be applied to a specific leader (see second operation of the above described algebraic vector quantization) to obtain a lattice point z; and
 c) if the quantized subvector B_{k }(a lattice point in lattice RE_{8}) was not in the base codebook, the 8 indices of the Voronoi extension index vector k calculated in suboperation V1 of the second operation of the above described algebraic vector quantization; from the Voronoi extension indices, an extension vector v can be computed as taught by Reference [8]. The number of bits in each component of index vector k is given by the extension order r, which can be obtained from the code value of index n_{k}. The scaling factor M of the Voronoi extension is given by M=2^{r}.

[0255]
Then, from the scaling factor M, the Voronoi extension vector v (a lattice point in lattice RE_{8}) and the lattice point z in the base codebook (also a lattice point in lattice RE_{8}), each quantized scaled subvector {circumflex over (B)}_{k }can be computed using the following relation:

[0000]
{circumflex over (B)}
_{k}
=Mz+v.

[0256]
When there is no Voronoi extension (i.e. n_{k}<5, M=1 and z=0), the base codebook is either codebook Q_{0}, Q_{2}, Q_{3 }or Q_{4 }from Reference [6]. No bits are then required to transmit vector k. Otherwise, when Voronoi extension is used because {circumflex over (B)}_{k }is large enough, then only Q_{3 }or Q_{4 }from Reference [6] is used as a base codebook. The selection of Q_{3 }or Q_{4 }is implicit in the codebook number value n_{k}, as described in the second operation of the above described algebraic vector quantization.

[0257]
Operation 908: An adder sums the firststage approximation from Operation 905 to the inverseweighed AVQ contribution from Operation 907 to reconstruct and recover the quantized LSF vector 909.

[0258]
Although the present invention has been defined in the foregoing description by means of illustrative embodiments thereof, these embodiments can be modified at will, within the scope of the appended claims, without departing from the spirit and nature of the present invention.
REFERENCES

[0000]
 [1] 3GPP Technical Specification TS 26.290, “Audio Codec Processing Functions; Extended Adaptive MultiRateWideband (AMRWB+) Codec; Transcoding Functions,” June 2005.
 [2] J. Skoglund, J. Linden, “Predictive VQ for Noisy Channel Spectrum Coding: AR Or MA?,” IEEE 1997 International Conference on Acoustics, Speech, and Signal Processing (ICASSP'97), pp. 13511354, Munich, Germany, Apr. 2124, 1997.
 [3] H. Zarrinkoub, P. Mermelstein, “Switched Prediction and Quantization of LSP Frequencies,” IEEE 1996 International Conference on Acoustics, Speech, and Signal Processing (ICASSP'96), Vol. 2, pp. 757760, 710 May 1996.
 [4] A. V. McCree, “Method for SwitchedPredictive Quantization,” U.S. Pat. No. 6,122,608.
 [5] R. Laroia, N. Phamdo, and N. Farvardin, “Robust and Efficient Quantization of Speech LSP Parameters Using Structured Vector Quantizers,” IEEE Int. Conf. on Acoustics, Speech and Signal Processing (ICASSP'1991), pp. 641644, Washington, D.C., Apr. 1417, 1991.
 [6] M. Xie and J.P. Adoul, “Embedded Algebraic Vector Quantization (EAVQ) with Application to Wideband Audio Coding,” IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), Atlanta, Ga., U.S.A., Vol. 1, pp. 240243, 1996.
 [7] P. Rault, C. Guillemot, “Indexing Algorithm for Zn, An, Dn and Dn++ Lattice Vector Quantizers,” IEEE Transactions on Multimedia, Vol. 3, No. 4, December 2001.
 [8] J. H. Conway and N. J. A. Sloane, “A Fast Encoding Method for Lattice Codes and Quantizers,” IEEE Trans. Inform. Theory, Vol. IT29, No. 6, pp. 820824, November 1983.